ASm

RVa

b

c

Ld

a

ARRAA

KIAOACM

1

umrihs

If

h0

Journal of Chromatography A, 1361 (2014) 67–76

Contents lists available at ScienceDirect

Journal of Chromatography A

j o ur na l ho me page: www.elsev ier .com/ locate /chroma

queous two-phase systems for enhancing immunoassay sensitivity:imultaneous concentration of mycotoxins and neutralization ofatrix interference

.R.G. Soaresa,c, P. Novoa,d, A.M. Azevedob,c, P. Fernandesb,c,

. Chua, J.P. Condea,b, M.R. Aires-Barrosb,c,∗

INESC Microsistemas e Nanotecnologias and IN-Institute of Nanoscience and Nanotechnology, Lisbon, PortugalDepartment of Bioengineering, Instituto Superior Técnico, Universidade de Lisboa, Lisbon, PortugalIBB—Institute for Biotechnology and Bioengineering, Centre for Biological and Chemical Engineering, Instituto Superior Técnico, Universidade de Lisboa,isbon, PortugalLeibniz-Institut für Analytische Wissenschaften - ISAS - e.V., Otto-Hahn-Str. 6b, 44227 Dortmund, Germany†

r t i c l e i n f o

rticle history:eceived 14 February 2014eceived in revised form 2 June 2014ccepted 1 August 2014vailable online 12 August 2014

eywords:mmunoassayqueous two-phase systemchratoxin Aflatoxin B1oncentrationatrix

a b s t r a c t

Immunoassays have a broad application range, from environmental and food toxicology to biomedicalanalysis, providing rapid and simple methods for analyte quantification. Immunoassays, however, areoften challenging at nM and sub nM concentrations and are affected by detrimental matrix interfer-ence effects, as is the case of the detection of ochratoxin A (OTA) and Aflatoxin B1 (AFB1). These arewidespread mycotoxins found in food and feed, with serious potential implications for human health.This work demonstrates the use of polymer–salt aqueous two phase systems (ATPSs) for the simulta-neous concentration of mycotoxins and neutralization of matrix interference. In particular, polyethyleneglycol (PEG)-phosphate salt ATPSs were used to enhance the detection sensitivity of OTA and AFB1 inwines and beer by an indirect competitive ELISA. Using this methodology it was possible to quantifyboth analytes spiked in red wine with limits-of-detection (LoD) down to 0.19 ng/mL and 0.035 ng/mL,respectively, with results comparable to those obtained using solutions of toxins in phosphate bufferedsaline (PBS) buffer (0.7 ng/mL and 0.009 ng/mL, respectively). Furthermore, a very low matrix-to matrix

variability was observed, with LoD and half inhibitory concentration (IC50) values of 5.17 ± 1.08 and33.2 ± 3.5 ng/mL (±SD) obtained in the detection of OTA spiked in red and white wines, beer or PBSbuffer. These results indicate the potential of ATPS as a fast and simple concentration step and in provid-ing matrix-independent analyte quantification for enhanced immunoassay sensitivity below regulatorylevels.

. Introduction

Immunoassays are powerful bioanalytical techniques that aresed extensively in medical diagnostics [1], allowing a simplerultiplexing using a different label for each target analyte (label

esolved methods) [2] or an array of spots, with each spot target-

ng a given analyte (space resolved methods) [3,4]. Immunoassaysave also been successfully broadened to environmental analy-is [5,6], with a significant and increasing share being targeted to

∗ Corresponding author at: Universidade de Lisboa, Department of Bioengineering,nstituto Superior Técnico, 1049-001 Lisboa, Portugal. Tel.: +351 218 419 134;ax: +351 218 419 062.

E-mail address: rabarros@tecnico.ulisboa.pt (M.R. Aires-Barros).† Current Address.

ttp://dx.doi.org/10.1016/j.chroma.2014.08.007021-9673/© 2014 Elsevier B.V. All rights reserved.

© 2014 Elsevier B.V. All rights reserved.

mycotoxin quantification in feeds and foodstuff [7–9]. However,immunoassays suffer from significant matrix interference prob-lems in samples such as blood [10], urine [11], water [5] and foodextracts [12–15]. This interference is responsible for a pronounceddecrease in sensitivity and inter-assay reproducibility.

Two mycotoxins, ochratoxin A (OTA) and aflatoxin B1 (AFB1)were selected as environmentally relevant model analytes fortesting with immunoassays. Both are of high importance due toconfirmed multiple toxic effects in animals and potential toxiceffects in humans such as carcinogenicity and hepatotoxicity[16,17], together with their widespread prevalence in foods andfeeds [18]. The detection of OTA and AFB1 using competitive

immunoassays has been well explored in literature [19–25] and isgaining an increasing amount of interest due to rapid assay timesand low-cost when compared to conventional chromatographicmethods [26,27]. This detection is particularly relevant in both

6 omato

wf[egtteipifcplh

fiAwabwootslppattntgdvwwti

2

2

N((t(Ats1tAApawp

8 R.R.G. Soares et al. / J. Chr

ines and beer [28–30], with regulatory limits set as low as 2 ng/mLor OTA in wines and 2 ng/mg for AFB1 in cereal based products31]. However, these matrices result in pronounced interferenceffects, resulting in a sharp decrease in analysis sensitivity. This isenerally related to the high concentration of polyphenols such asannins, anthocyanins and gallic acid [32–34], which are reportedo reversibly or irreversibly bind to proteins [35,36]. Hence, thexisting immunoassays for mycotoxin quantification in wines, fornstance, are often dependent on some type of matrix clean-uprocedure such as solid phase extraction [20,21] or matrix neutral-

zation by the addition of NaHCO3 and polyethylene glycol (PEG),or example [22]. While the former provides simultaneous matrixlean-up and analyte concentration, it increases the costs and com-lexity of the assay. On the other hand, while being simple, the

atter implies sample dilution, which is obviously undesirable whenigh sensitivity is demanded.

In this paper, we use a simple aqueous two phase system (ATPS)or rapid and low-cost analyte concentration and matrix neutral-zation, prior to performing an indirect competitive ELISA (icELISA).TPS is a type of liquid–liquid separation phenomenon that occurshen certain compounds are mixed in an aqueous solution above

particular critical concentration [37]. This solution is then capa-le of splitting into two or more spontaneously generated phases,ith different chemical compositions, but both comprised mainly

f water. There are many pairs of such compounds that are capablef generating biphasic interfaces like polymer–polymer [38] sys-ems, polymer–salt systems [39], thermosensitive polymer–waterystems [40], alcohol-salt systems [41] and polymer- or salt-ioniciquid systems [42,43]. Among these, the PEG-NaH2PO4/K2HPO4air, which has been widely used by our research group for proteinurification [44–46], is chosen here as the model matrix clean-upnd toxin concentration system. The choice of this ATPS pair allowshe exploitation of the high electrochemical gradient between thewo water-rich phases, inherent to this system, resulting in a pro-ounced uneven molecular partition behavior [47]. This allowshe optimization of conditions to effectively concentrate a tar-et analyte by manipulating the phase volumes, by increasing orecreasing the final volume ratio (Vr) of the system, defined as theolume of top phase divided by the volume of bottom phase. Thus,e propose a novel combination of PEG-salt ATPS-based extractionith a well plate icELISA for simultaneous analyte concentra-

ion and screening of matrix interference, leading to enhancedmmunoassay performance.

. Experimental methods

.1. Materials

PEG with 2, 6 or 10 kDa average molecular weight (MW),aH2PO4, K2HPO4, OTA, AFB1, phosphate buffered saline

PBS) tablets, methanol (99.9%), OTA-bovine serum albuminBSA) conjugate, AFB1-BSA conjugate, Tween 20, 3,3′,5,5′-etramethylbenzidine (TMB) ready-to-use liquid substratesuper-slow) for ELISA and BSA were purchased from Sigmaldrich. Ultrapure water was obtained from a MilliQ purifica-

ion system from Millipore (Billerica, MA, USA). OTA and AFB1tock solutions were prepared with an equal concentration of00 �g/mL in 25% (v/v) and 50% (v/v) methanol in PBS, respec-ively. The anti-OTA IgG and anti-AFB1 IgG were purchased frombcam (Cambridge, UK) as monoclonal mouse antibodies (3C5 andFA-1, respectively). Horseradish peroxidase (HRP) conjugated

olyclonal goat anti-mouse IgG (whole molecule) was purchasedlso from Abcam. Generic Portuguese red wine (11% v/v ethanol),hite wine (11% v/v ethanol) and beer (5.2% v/v ethanol), wereurchased from a local store.

gr. A 1361 (2014) 67–76

2.2. ATPS preparation for evaluation of analyte partition

The ATPSs were prepared for a total of 1 g in 1.5 mL microtubes.Aqueous solutions of PEG 50% (w/w) (2000, 6000 or 10,000 Daaverage MW and NaH2PO4/K2HPO4 40% (w/w) (adjusted to a finalpH of 6, 7 or 8 by changing the ratio of monosodic to dipotas-sic salts added) were added depending on the conditions chosenand adjusted to the total mass of the system using ultrapure water.Then, each tube was agitated in a vortex for 10 to 15 s. Finally, toobtain clear separated phases the microtubes were centrifuged at6000 × g for 2 min. The analytical balance used was a Mettler Toledo(Columbus, OH, USA) XS205 Dual Range. The statistical analysis ofthe results was performed using the computer software Statistica7 from StatSoft (Tulsa, OK, USA).

2.3. Fluorimetric quantification of the analytes

Analyte quantification in both phases of 1 �g/mg analyte spikedATPS was performed using a Varian (Palo Alto, CA, USA) Cary Eclipseplate reader spectrofluorimeter. All experiments were performedusing a high voltage (800 V) setting for the photomultiplier tube andthe excitation/emission slits were set at 5 nm. The readings wereaveraged from 5 consecutive measurements with a 0.5 s intervalin white polystyrene Corning® (Corning, NY, USA) 96 well plates.The excitation and emission wavelengths for OTA and AFB1 quan-tification were set at 333/446 nm and 370/439 nm, respectively. Asingle point calibration of the fluorescence emission was performedby adding a known amount of analyte (2.5 �g/mL) to both top andbottom phases of a second ATPS prepared for each condition. Thisaddition was performed after separating each phase to individ-ual microtubes. A third ATPS was also prepared to subtract thebackground fluorescence value for each phase. The validity of thisapproach is supported by the linear dependence of the fluorescenceemission as a function of the analyte concentration (Fig. A-1).

2.4. ATPS preparation for analyte immunodetection

Undiluted red wine, white wine and beer were first adjustedto pH 6 using 10 M NaOH before any further processing. ThenNaH2PO4/K2HPO4 were added at a mass ratio of 1.27 or 0.43 fora final pH of 6 or 7, respectively, and weighed together with wineor beer to the intended final mass concentration. This mixture wasthen agitated in a vortex for 20–30 s for complete salt dissolution.Then, 1881 �L of this solution were transferred to individual 2 mLmicrotubes and spiked with 19 �L of an OTA or AFB1 solutions pre-pared in PBS, at a concentration 100 times higher than the intendedfinal concentration in wine. For the reference solutions, 19 �L ofPBS were added instead of the OTA or AFB1. A 50% (w/w) aque-ous solution of the second phase forming component (PEG 1000,8000 or 20,000) was then added, 4–5 mg at a time, with intermit-tent agitations for 20–30 s in a vortex and 2 min of centrifugationat 6000 × g. This process was stopped when a collectable amountof top phase was obtained and was only performed the first time anew system was used or a new top phase volume was required. Thecollected top phase was then diluted with a PBS solution contain-ing anti-OTA or anti-AFB1 IgG, yielding the final IgG concentrationof 50 or 100 ng/mL, respectively. Depending on the experiment, incases where polyphenol precipitation occurred after the dilutionstep, the solution was centrifuged a second time under the abovementioned conditions. This final solution was then used as SUA(sample under analysis) for the wellplate icELISA.

2.5. Indirect competitive ELISA

MaxiSorp® flat-bottom 96 well microtiter plates from Nunc(Roskilde, Denmark) and Bio-one PS medium bind ELISA plates from

R.R.G. Soares et al. / J. Chromatogr. A 1361 (2014) 67–76 69

F (b) ma st sign

GeFsiasT0tapoa1aap1ts

ig. 1. Molecular structure and K surface tendency charts of each OTA (a) and AFB1sterisk has a pKa of 7.1. The factors plotted in each tendency chart provide the mo

reiner (Kremsmünster, Austria) were used for the OTA and AFB1xperiments, respectively. The shaker-incubator used was a Stat-ax 2200 incubator/shaker from Bio-Rad (Hercules, CA, USA) set at ahaking speed of 5. First, 100 �L of a 10 �g/mL solution of OTA-BSAn carbonate buffer (100 mM) at pH 9.2 or AFB1-BSA in PBS weredded to each well. The plate was then incubated at 37 ◦C withhaking for 1 h. After this coating step, the solution was discarded.hen, each well was washed twice with 100 �L of PBS containing.1% (v/v) Tween 20 (PBST). In each washing step, the plate was agi-ated for 30 s and the liquid was discarded. After the second washnd removal of the liquid, 150 �L of a 0.5% (w/v) BSA solution pre-ared in PBS was added to the wells to block the uncoated areasn the polystyrene surface. This solution was incubated for 30 mint room temperature (RT) with shaking. After washing three times,00 �L of the SUA, containing 50 ng/mL anti-OTA IgG or 100 ng/mLnti-AFB1 IgG, were added to the wells and incubated for 45 mint RT with shaking. Four washing steps with PBST followed the

rimary antibody step. The final step comprised the addition of00 �L of a secondary antibody labeled with HRP, at a concentra-ion of 100 ng/mL in PBST. The plate was incubated for 30 min withhaking at RT. Four final washing steps were performed also with

olecules in PEG-NaH2PO4/KH2PO4 ATPSs. The OTA phenol moiety labeled with anificant effect on the variable K (Fig. A-2).

PBST and the plate was tapped on blotting paper to ensure max-imum removal of the remaining washing solution. Finally, 100 �Lof a commercial TMB solution, containing hydrogen peroxide, wereadded to the wells, followed by 1–5 min incubation at RT withoutshaking, depending on the time taken for color development. Thereaction was stopped by adding 100 �L of 250 mM HCl solution.The plate was then inserted into the spectrophotometer within5–10 min of stopping the reaction and the absorbance was mea-sured at a wavelength of 450 nm. The absorbance value obtainedfor each SUA is normalized in respect to a reference solution withthe same matrix and IgG composition but without OTA or AFB1spiking.

3. Results and discussion

3.1. Partition behavior of OTA and AFB1 in ATPS

In order to develop a clean-up and concentration strategy basedon ATPS, the partition coefficient (K) of the target analytes was firstanalyzed. The molecular structure of both OTA and AFB1 is pre-sented in Fig. 1(a) and (b). A face centered central composite design

70 R.R.G. Soares et al. / J. Chromatogr. A 1361 (2014) 67–76

Fig. 2. Schematics of the ATPS mediated concentration strategy and main optimization parameters, highlighted in red. (a) The gray arrow refers to phase separation aftercentrifugation. The cylinder represents a well after previous adsorption of analyte-BSA conjugate. (b) Photograph of a tube after addition of 55 mg of a PEG 20000 (50% w/w)s issolvl

oelst[fPftafl

a(trtoat

olution to red wine in which 12.5% (w/w) of NaH2PO4/K2HPO4 were previously degend, the reader is referred to the web version of this article.)

f experiments (DoE) was used to evaluate three key variables forach analyte [49–52], which are the pH, PEG MW and the tie-lineength (TLL). Briefly, the TLL represents the variation in the compo-ition of the top PEG-rich and bottom salt-rich phases and the largerhe difference in the composition of the phases, the longer the TLL37]. The TLL values were considered as approximate unitary valuesrom 1 to 3, which correspond to equal and linearly increasing totalEG and salt concentrations which is a reasonable trend indicationor the K values, and avoids the time-consuming quantification ofhe PEG and salt mass fractions in each of the phases. The resultsre summarized in Fig. 1(a) and (b). Details on the conditions usedor each K measurement and respective extraction yield values areisted on Table A-1.

For the same set of factors and conditions tested, tenfold higherverage values of K are obtained for OTA (K ≈ 267) than for AFB1K ≈ 22), while in both cases the average extraction yields are closeo 100% in the top PEG rich phase (Table A-1). Several literatureeports claim a good correlation between molecular partition to

op phase in PEG-salt ATPS and increasing experimental/calculatedctanol–water partition coefficient (log POW) values [53–56], whichrises from the intrinsic difference in hydrophobicity betweenhe more hydrophilic salt-rich phase (water-like), and the more

ed, for a final pH of 7. (For interpretation of the references to color in this figure

hydrophobic PEG-rich phase (octanol-like). These reports and thehigher K obtained for OTA correlate with the fact that OTA and AFB1have calculated [57] log POW values of 4.61 and 1.58, respectively.However, at the pH values tested, the major OTA microspecies haveeither a mononegative or dinegative charge, with the calculated[57] pH dependent log POW within 2 to 1, while AFB1 remains neu-tral. Thus, a significant difference in lipophilicity does not seem toexist between both molecules in the pH range tested. Alternatively,the difference in K values may possibly be related with the inher-ent positive electrochemical potential generated in the top phase ofPEG-phosphate systems [47]. To maintain electroneutrality at equi-librium, the system might favor the partition of negatively chargedmolecules to the top phase. Such hypothesis is supported by theresults obtained by Bora and co-workers, regarding the partitionof the cephalosporins 7-ADCA and cephalexin in multiple PEG-saltsystems [58]. The overall electric charge of these molecules shiftswith pH, in a given range of pH [57]. Accordingly, the partition grad-ually shifts from the more hydrophilic salt phase to the PEG phase,

as observed here for the mono/dinegative OTA, in opposition to theneutral AFB1.

OTA partition displays a very strong dependence on the TLLand a weaker, but significant (p < 0.05), dependence on the pH (for

R.R.G. Soares et al. / J. Chromatogr. A 1361 (2014) 67–76 71

Fig. 3. Schematics of the well plate icELISA used for OTA quantification in the absence (top, high signal) or presence (bottom, low signal) of OTA in solution (a). I—Analyte-BSA conjugate adsorption; II—addition of the sample under analysis; III—addition of a HRP labeled secondary antibody; IV—addition of TMB solution containing hydrogenperoxide; V—addition of an acidic (250 mM HCl) solution. The blue arrows represent washing steps with PBST. (b) Main binding kinetic parameters associated with thei h the

l

smmbiceoctKpi

ndirect competitive assay principles and hypothetical interfering interactions witegend, the reader is referred to the web version of this article.)

tandardized effect charts, refer to Fig. A-2). A longer TLL as a pro-oter of more extreme K values has been reported [47,53,59]. Theore pronounced difference in phase composition may enhance

oth the electrochemical and polarity gradients present at thenterface, thus further emphasizing the entropic and enthalpicontributions for molecular partition [47]. The pH effect may bexplained by the increasing concentration of the dinegative formf OTA in the pH range tested [57], in tune with the previously dis-ussed hypothesis that the negative charge promotes the partition

o the PEG rich phase. As for AFB1, the TLL factor weakly affects the

value, (Fig. A-2 and Fig. 1(b)), while pH has no significant effect onartition, which is expected as AFB1 remains with a neutral charge

n the tested pH range.

polyphenol molecules. (For interpretation of the references to color in this figure

Thus, our system is neither significantly sensitive to tempera-ture, nor requires the use of additional molecules to enhance the Kof the target analytes, unlike results previously reported involvingthe use of water-surfactant systems to concentrate proteins andvirus to enhance detection in lateral flow immunoassays [60,61].Furthermore, the separation times are reduced from several hoursto a few minutes.

3.2. Optimization of the ATPS mediated concentration to enhance

the sensitivity of an icELISA

Both analytes partition strongly to the top PEG rich phase, thusan ATPS concentration strategy (Fig. 2(a)) was developed, based on

72 R.R.G. Soares et al. / J. Chromatogr. A 1361 (2014) 67–76

Fig. 4. Effect of salt/PEG related parameters and top phase dilution on the icELISA LoD and IC50 values obtained for OTA quantification. For the dilution effect plot, the valueson the x-axis are calculated as the volume fraction of top phase per total volume of final solution. Red wine was used as SUA for all experiments. Refer to supporting info ford and rab of eacr o colo

ttfPtTisiacip

etails regarding the remaining conditions used for each factor analysis (Table A-2)

etween 90 and 99% (*) and above 99% (**), for the hypothesis of the average valueefer to the LoD and IC50 values, respectively. (For interpretation of the references t

he generation of systems with a small Vr. Under these conditionshe colored compounds in wine and beer, which include a largeraction of the polyphenols [32], tend also to concentrate on theEG rich phase, as can be visualized in Fig. 2(b) Possible interac-ions due to the presence of polyphenols are shown in Fig. 3(b).o quantify the amount of the target analytes in solution using ancELISA (Fig. 3(a)), the increasing concentration of analyte needs tohift the equilibrium toward the free antibody through an increasen k1, which in turn decreases the amount of antibody bound to the

dsorbed analyte-BSA conjugate, promoted by an approximatelyonstant kon (assuming a reproducible surface coverage). Interfer-ng molecules in solution such as polyphenols can possibly bindroteins closer to key sites such as the analyte in the surface

w curve data (Fig. A-7). The asterisk code refers to a confidence level below 90% (o),h sample group being independent. The symbols on the left (black) or right (blue)r in this figure legend, the reader is referred to the web version of this article.)

adsorbed BSA or at close proximity to the antibody binding site, pos-ing a direct (k1-PP) or indirect (kon-PP) stereochemical hindrance,respectively. It was observed that while the red wine polyphenolsinteract in fact with both BSA and IgG-FITC conjugates, resulting ina quenching effect on the FITC fluorescence, the presence of PEGrich phase (Fig. A-3(a)) allows the reduction of such interactions,resulting in a pronounced increase in fluorescence emission (Fig.A-3(b)). This preferential affinity of the polyphenols to PEG [62],responsible for reducing the effect of interference on immunoas-

says has been reported [14,21,22]. Still, concentrations of PEG equalor lower than 5% (w/w) did not provide a significant enhancementof the icELISA analyte detection performance, when added directlyto raw red wine, in comparison to the results in PBS (Fig. A-4).

R.R.G. Soares et al. / J. Chromatogr. A 1361 (2014) 67–76 73

Fig. 5. Comparison of icELISA sensitivity using either raw red/white wines or beer matrices as SUA (top) or ATPS processed samples (bottom). The curve obtained in PBS bufferi lues aa re ant

abMaaarsa

otTfsPosur

S

s represented in both conditions for direct visual comparison. The relative signal vand the average absorbance value of the blank (0 ng/mL analyte). All blank values awo independent assays.

Since the icELISA requires both the target analyte and annalyte-specific antibody to be present in solution, the partitionehavior of IgG was assessed under different PEG concentrations,W and salt concentrations (Fig. A-5). Depending on the PEG MW

nd TLL present in the system, the IgG molecules partition prefer-bly to either of the phases or precipitate at the interface above

certain maximum TLL threshold, in agreement with previousesults [44,46,63]. Accordingly, the antibody was added after phaseeparation together with the diluting solution, in order to provide

reproducible amount of IgG in the PEG rich phase solution.The highest detection interference for both analytes was

bserved for immunodetection in red wine, with a negligible signalo noise below 100 ng/mL of analyte and a very low absolute signal.his matrix was thus selected as a worst case scenario and was usedor further ATPS concentration optimization studies. The effect ofalt mass concentration (MS), final pH (pHS), mass (MP) and MWP ofEG and the dilution of top PEG rich phase with IgG solution (DTP)n the LoD and IC50 for OTA was assessed and the results are pre-ented in Fig. 4. The obtained LoD and IC values were determined

50sing the logarithmic fit shown as Eq. (1), with SR being the signalatio obtained at a given analyte concentration [A].

R = m × ln ([A]) + b (1)

re calculated as the ratio between the absorbance at a given analyte concentration average of four independent assays, while the remaining values are an average of

From Fig. 4 it can be concluded that even in non-optimal ATPSconditions, higher sensitivities were obtained in comparison to theaddition of PEG to the SUA (Fig. A-4), as previously reported [22].MS does not affect significantly neither the LoD nor the IC50 at con-centrations up to 18% (w/w), but at higher concentrations of 25%,the detection performance is severely impaired. On the other hand,the pHS has a statistically significant impact on the IC50 only, in therange of tested values. The decrease of MP, which defines the Vr

of the system and therefore the extent of the concentration effect,seems to have a positive impact on both the IC50 and LoD. Finally,the MWP, together with the DTP seem to have the greatest impact inboth the LoD and IC50 values obtained. While the MWP provides asignificant improvement in both variables at higher values, the DTP

seems to have an optimum range between 0.1 and 0.4 parts PEGrich phase in 0.9 to 0.6 parts diluting solution.

The extent of PEG concentration in the PEG rich phase thus playsa key role in preventing a further increase in detection sensitivity.This is reinforced by the similar results obtained for the same ATPSconditions applied to a wine matrix or PBS buffer (Fig. A-2). Further-

more, the results also strongly suggest that increasing PEG MWsprovide different interactions with BSA and/or IgG proteins, result-ing in dramatic differences in sensitivity. Reports in the literature

74 R.R.G. Soares et al. / J. Chromatogr. A 1361 (2014) 67–76

Table 1Summary of results obtained for the optimized ATPE-icELISA conditions.

Analyte Matrix MWP (kDa) LoDa (ng/mL) IC50a (ng/mL) −ma

OTA PBS – 0.70 9.50 17.74OTA RWb – >100 >100 0.52OTA WWc – 45.73 >100 8.22OTA BRd – 9.72 73.94 19.70OTA RW 8 6.37 35.40 23.88OTA WW 8 4.06 32.57 21.54OTA BR 8 5.77 36.38 24.01OTA RW 20 0.19 6.27 9.12OTA WW 20 0.17 10.27 11.20OTA BR 20 1.77 7.39 11.71AFB1 PBS – 0.009 0.15 15.08AFB1 RW – >100 >100 <0AFB1 WW – 3.273 4.10 9.26AFB1 BR – 0.436 8.56 12.23AFB1 RW 8 0.092 0.91 19.45AFB1 WW 8 0.176 0.71 25.88AFB1 BR 8 0.143 0.84 25.80AFB1 RW 20 0.035 1.02 11.18AFB1 WW 20 0.043 1.09 8.50AFB1 BR 20 0.021 0.95 10.74

a Calculated with Eq. (1).b

atatr[iAscutPc

apte

sAOalssmrscPstbte1wMl

Fig. 6. icELISA results for AFB1 spiked in red wine, processed with ATPS with increas-ing MS values. The remaining conditions are: pHS = 7; MP = 60 mg; MWP = 8 kDa;DTP = 0.2 (a) or pHS = 7; MP = 60; MWP = 20 kDa; DTP = 0.3 (b). The relative signal valuesare calculated as the ratio between the absorbance at a given analyte concentration

Red wine,c White wine.d Beer.

re contradictory on this topic. While some authors have concludedhat volume exclusion effects predominate and that protein inter-ction with PEG is negligible [64], Wu and co-workers, reported thathere is a significant interaction of BSA with PEG with increasingatios of protein to PEG, with a non-linear dependence on the MW65]. This non-linear dependence was also observed when measur-ng the polyphenol mediated quenching of FITC labeled IgG (Fig.-6), in the presence of 1% of PEG with different MWs. The resultsuggest a low point in polyphenol binding using PEG 8000. The TMBonversion rate was also considerably higher when PEG 8000 wassed, in comparison to 1000 or 20,000 (Fig. A-6), which supportshe previous observations. Such non-linear dependence, regardingEG-tannin binding in particular, was also observed by Makkar ando-workers [66].

While the generation of a smaller volume of PEG rich phasepparently provides lower IC50 values, the increase of the LoD isrobably due to the practical limitation of collecting, without con-amination with bottom salt rich phase, a larger fraction of thexistent PEG rich phase.

Using a set of conditions directed toward increased detectionensitivity, the icELISA results from raw matrices, PBS buffer andTPS concentrated solutions were compared. The results for bothTA and AFB1 are presented in Fig. 5 and the calculated LoD, IC50nd slope values for each of the curves are listed in Table 1. Regard-ess of the matrix used, the quantification performance was veryimilar for each MWP. PEG 20000 in particular, allowed for a veryensitive detection and a linear range spanning three orders ofagnitude were achieved. PEG 8000 prevented detection at the

egulatory limits, yet the response is more accurate, with a higherlope, as shown in Fig. A-2, covering a smaller range of analyte con-entrations. Furthermore, the analyte concentration effect usingEG 20000 was sufficient to provide a superior OTA detection sen-itivity in red wine, when compared to PBS control conditions, buthe same was not achieved for AFB1. Such difference can possiblye related to the lower K values inherent to AFB1 partition to theop PEG rich phase. Accordingly, and given the previously observedffect of the TLL on increasing the K values, MS values of 15% and

7.5% NaH2PO4/K2HPO4 were also tested (Fig. 6). Unexpectedly,hen using PEG 8000, the obtained curves overlap regardless of theS. Also, in conditions where PEG 20000 is used, there is a slight

oss of linearity in the log scale with increasing salt concentration,

and the average absorbance value of the blank (0 ng/mL analyte). All blank valuesare an average of four independent assays, while the remaining values were theaverage of two independent assays.

resulting in an increase of LoD from 0.035 ng/mL to 0.098 ng/mL anda decrease in IC50 from 1.02 ng/mL to 0.67 ng/mL, from 12.5% MS to15% MS. This may be either due to an increased PEG 20000 concen-tration in the PEG rich phase, as a consequence of an increase in theTLL, or it may alternatively be related to a simultaneously increasedconcentration of polyphenols. In either case, these results are inaccordance with the above mentioned observation that an MS upto 18% does not have a significant impact in detection performance.

4. Conclusions

This work demonstrated the successful use of polymer–saltATPS to enhance analyte detection sensitivity in immunoassays,allowing a simultaneous concentration and neutralization of matrixinterference. Using a model icELISA to quantify two environmen-tally relevant mycotoxins in complex wine and beer matrices,the results were improved at least by one order of magnitude

in comparison to the detection in raw matrices, providing eithercomparable or better results than those obtained using PBS con-trol conditions. The MWP had a significant impact on the detectioncurve. This allowed for either a more sensitive but less accurate

omato

q(c(vs

ifpptpa[isoTuoomd

A

(pEoF

A

i2

R

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

R.R.G. Soares et al. / J. Chr

uantification, which spans a large analyte concentration rangeMWP = 20 kDa) or a less sensitive, but more accurate quantifi-ation which can be directed toward “yes-no” types of responseMWP = 8 kDa). The developed method proved to be robust in pro-iding a high and similar sensitivity regardless of the analyte beingpiked in PBS or polyphenol-rich wine.

An ATPS based strategy to improve immunodetection sensitiv-ty might be easily broadened to other relevant mycotoxins, viz.umonisins, zearalenone and tricothecenes, in food or feed sam-les. This can be done by optimizing ATPS conditions, using knownhase forming compounds [67,68], and combining analyte concen-ration and sample clean-up. The development of simple samplerocessing steps, such as the one presented, is gaining relevances regulatory limits are including an increased number of toxins69] and their limiting values tend to decrease [70]. Furthermore,t is expected that under optimized conditions this ATPS basedtrategy may also be broadened to improve the immunodetectionf clinically relevant biomarkers such as proteins and DNA/RNA.he pace at which new developments within this novel applicationnfold would benefit with robotic high throughput screening meth-ds [71], design of experiment approaches [50,72,73], applicationf genetic algorithms [74] or modeling of solvent properties andolecular dynamics approaches [48,75,76], that were previously

eveloped toward molecular purification using ATPS.

cknowledgements

Pedro Novo acknowledges FCT for his PhD grantSFRH/BD/74037/2010). This work was supported by Fundac ãoara a Ciência e a Tecnologia (FCT) through the Projectostratégico PEst-OE/CTM/LA0024/2013 of the Associated Lab-ratory IN—Institute of Nanoscience and Nanotechnology and theCT Research Project of Excellency EXCL/CTM-NAN/0441/2012.

ppendix A. Supplementary data

Supplementary material related to this article can be found,n the online version, at http://dx.doi.org/10.1016/j.chroma.014.08.007.

eferences

[1] H. Päkkilä, T. Soukka, Simple and inexpensive immunoassay-based diagnostictests, Bioanal. Rev. 3 (2011) 27–40.

[2] H. Li, Z. Cao, Y. Zhang, C. Lau, J. Lu, Combination of quantum dot fluorescencewith enzyme chemiluminescence for multiplexed detection of lung cancerbiomarkers, Anal. Methods 2 (2010) 1236.

[3] C. Zong, J. Wu, C. Wang, H. Ju, F. Yan, Chemiluminescence imaging immunoas-say of multiple tumor markers for cancer screening, Anal. Chem. 84 (2012)2410–2415.

[4] M. Lee, S. Lee, J.H. Lee, H.W. Lim, G.H. Seong, E.K. Lee, S.I. Chang, C.H. Oh, J.Choo, Highly reproducible immunoassay of cancer markers on a gold-patternedmicroarray chip using surface-enhanced Raman scattering imaging, Biosens.Bioelectron. 26 (2011) 2135–2141.

[5] A.A.d.l. Cruz, The effects of sample matrices on immunoassays to detectmicrocystin-LR in water, J. Environ. Prot. 03 (2012) 1275–1285.

[6] M. Oplatowska, C.T. Elliott, A.C. Huet, M. McCarthy, P.P. Mulder, C. von Holst,P. Delahaut, H.P. Van Egmond, K. Campbell, Development and validation of arapid multiplex ELISA for pyrrolizidine alkaloids and their N-oxides in honeyand feed, Anal. Bioanal. Chem. 406 (2013) 757–770.

[7] L. Anfossi, C. Baggiani, C. Giovannoli, G. D’Arco, G. Giraudi, Lateral-flowimmunoassays for mycotoxins and phycotoxins: a review, Anal. Bioanal. Chem.405 (2013) 467–480.

[8] E.N. Ediage, J.D. Di Mavungu, I.Y. Goryacheva, C. Van Peteghem, S. De Saeger,Multiplex flow-through immunoassay formats for screening of mycotoxins ina variety of food matrices, Anal. Bioanal. Chem. 403 (2012) 265–278.

[9] J. Peters, M. Bienenmann-Ploum, T. de Rijk, W. Haasnoot, Development of amultiplex flow cytometric microsphere immunoassay for mycotoxins and eval-uation of its application in feed, Mycotoxin Res. 27 (2011) 63–72.

10] J. Tate, G. Ward, Interference in immunoassay, Clin. Biochem. Rev. 25 (2004)105–120.

[

gr. A 1361 (2014) 67–76 75

11] T.P. Taylor, M.G. Janech, E.H. Slate, E.C. Lewis, J.M. Arthur, J.C. Oates, Overcomingthe effects of matrix interference in the measurement of urine protein analytes,Biomarker Insights 7 (2012) 1–8.

12] Z. Dzuman, M. Vaclavikova, I. Polisenska, Z. Veprikova, M. Fenclova, M.Zachariasova, J. Hajslova, Enzyme-linked immunosorbent assay in analysis ofdeoxynivalenol: investigation of the impact of sample matrix on results accu-racy, Anal. Bioanal. Chem. 406 (2014) 505–514.

13] X.H. Wang, T. Liu, N. Xu, Y. Zhang, S. Wang, Enzyme-linked immunosorbentassay and colloidal gold immunoassay for ochratoxin A: investigation of analyt-ical conditions and sample matrix on assay performance, Anal. Bioanal. Chem.389 (2007) 903–911.

14] M.M. Ngundi, L.C. Shriver-Lake, M.H. Moore, M.E. Lassman, F.S. Ligler, C.R. Taitt,Array biosensor for detection of ochratoxin A in cereals and beverages, Anal.Chem. 77 (2004) 148–154.

15] P. Novo, G. Moulas, D.M. Franc a Prazeres, V. Chu, J.P. Conde, Detection of ochra-toxin A in wine and beer by chemiluminescence-based ELISA in microfluidicswith integrated photodiodes, Sens. Actuators, B: Chem. 176 (2013) 232–240.

16] A.el Khoury, A. Atoui, Ochratoxin a: general overview and actual molecularstatus, Toxins 2 (2010) 461–493.

17] C.P. Wild, Y.Y. Gong, Mycotoxins and human disease: a largely ignored globalhealth issue, Carcinogenesis 31 (2010) 71–82.

18] I. Rodrigues, K. Naehrer, Mycotoxin survey report 2012 (part I), AllAboutFeed21 (2013) 6–8.

19] I.Y. Goryacheva, T.Y. Rusanova, N.V. Beloglazova, I.I. Voronov, S. Saeger, Deter-mination of ochratoxin A in colored food products: sample preparation and animmunoassay test method, J. Anal. Chem. 65 (2010) 760–766.

20] F. Zezza, F. Longobardi, M. Pascale, S.A. Eremin, A. Visconti, Fluorescence polar-ization immunoassay for rapid screening of ochratoxin A in red wine, Anal.Bioanal. Chem. 395 (2009) 1317–1323.

21] T.Y. Rusanova, N.V. Beloglazova, I.Y. Goryacheva, M. Lobeau, C. Van Peteghem,S. De Saeger, Non-instrumental immunochemical tests for rapid ochratoxin Adetection in red wine, Anal. Chim. Acta 653 (2009) 97–102.

22] L. Anfossi, C. Giovannoli, G. Giraudi, F. Biagioli, C. Passini, C. Baggiani, A lateralflow immunoassay for the rapid detection of ochratoxin A in wine and grapemust, J. Agric. Food Chem. 60 (2012) 11491–11497.

23] W. Jiang, Z. Wang, G. Nölke, J. Zhang, L. Niu, J. Shen, Simultaneous determi-nation of aflatoxin B1 and aflatoxin M1 in food matrices by enzyme-linkedimmunosorbent assay, Food Anal. Method 6 (2012) 767–774.

24] K.E. Sapsford, C.R. Taitt, S. Fertig, M.H. Moore, M.E. Lassman, C.M. Maragos, L.C.Shriver-Lake, Indirect competitive immunoassay for detection of aflatoxin B1in corn and nut products using the array biosensor, Biosens. Bioelectron. 21(2006) 2298–2305.

25] D. Zhang, P. Li, Y. Yang, Q. Zhang, W. Zhang, Z. Xiao, X. Ding, A high selectiveimmunochromatographic assay for rapid detection of aflatoxin B(1), Talanta85 (2011) 736–742.

26] F. Rubio, L.J. Veldhuis, B.S. Clegg, J.R. Fleeker, J.C. Hall, Comparison of a directELISA and an HPLC Method for glyphosate determinations in water, J. Agric.Food Chem. 51 (2002) 691–696.

27] A. Marcobal, M.C. Polo, P.J. Martın-Álvarez, M.V. Moreno-Arribas, Biogenicamine content of red Spanish wines: comparison of a direct ELISA and an HPLCmethod for the determination of histamine in wines, Food Res. Int. 38 (2005)387–394.

28] R. Mateo, A. Medina, E.M. Mateo, F. Mateo, M. Jimenez, An overview of ochra-toxin A in beer and wine, Int. J. Food Microbiol. 119 (2007) 79–83.

29] A. El Khoury, T. Rizk, R. Lteif, H. Azouri, M.L. Delia, A. Lebrihi, Fungal contami-nation and aflatoxin B1 and Ochratoxin A in Lebanese wine-grapes and musts,Food Chem. Toxicol. 46 (2008) 2244–2250.

30] B. Odhav, V. Naicker, Mycotoxins in South African traditionally brewed beers,Food Addit. Contam. 19 (2002) 55–61.

31] Commission Regulation (EC) No 1881/2006, Setting maximum levels for certaincontaminants in foodstuffs, Official Journal of European Union, 19 December2006, p. L364/315–316.

32] B. Sun, M.C. Leandro, V. de Freitas, M.I. Spranger, Fractionation of red winepolyphenols by solid-phase extraction and liquid chromatography, J. Chro-matogr. A 1128 (2006) 27–38.

33] C. Alcalde-Eon, M.T. Escribano-Bailón, C. Santos-Buelga, J.C. Rivas-Gonzalo,Changes in the detailed pigment composition of red wine during maturity andageing, Anal. Chim. Acta 563 (2006) 238–254.

34] S. Gorinstein, A. Caspi, M. Zemser, S. Trakhtenberg, Comparative contents ofsome phenolics in beer, red and white wines, Nutr. Res. 20 (2000) 131–139.

35] T. Ozdal, E. Capanoglu, F. Altay, A review on protein–phenolic interactions andassociated changes, Food Res. Int. 51 (2013) 954–970.

36] A. Bennick, Interaction of plant polyphenols with salivary proteins, Crit. Rev.Oral. Biol. Med. 13 (2002) 184–196.

37] S. Raja, V.R. Murty, V. Thivaharan, V. Rajasekar, V. Ramesh, Aqueous two phasesystems for the recovery of biomolecules—a review, Sci. Technol. 1 (2012) 7–16.

38] P.A. Rosa, A.M. Azevedo, I.F. Ferreira, S. Sommerfeld, W. Backer, M.R. Aires-Barros, Downstream processing of antibodies: single-stage versus multi-stageaqueous two-phase extraction, J. Chromatogr. A 1216 (2009) 8741–8749.

39] A.M. Azevedo, A.G. Gomes, P.A.J. Rosa, I.F. Ferreira, A.M.M.O. Pisco, M.R. Aires-Barros, Partitioning of human antibodies in polyethylene glycol–sodium citrate

aqueous two-phase systems, Sep. Purif. Technol. 65 (2009) 14–21.

40] P.L. Show, C.P. Tan, M. Shamsul Anuar, A. Ariff, Y.A. Yusof, S.K. Chen, T.C.Ling, Extractive fermentation for improved production and recovery of lipasederived from Burkholderia cepacia using a thermoseparating polymer in aque-ous two-phase systems, Bioresour. Technol. 116 (2012) 226–233.

7 omato

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

6 R.R.G. Soares et al. / J. Chr

41] C.W. Ooi, B.T. Tey, S.L. Hii, S.M.M. Kamal, J.C.W. Lan, A. Ariff, T.C. Ling, Purifica-tion of lipase derived from Burkholderia pseudomallei with alcohol/salt-basedaqueous two-phase systems, Process Biochem. 44 (2009) 1083–1087.

42] P. Berton, R.P. Monasterio, R.G. Wuilloud, Selective extraction and determina-tion of vitamin B12 in urine by ionic liquid-based aqueous two-phase systemprior to high-performance liquid chromatography, Talanta 97 (2012) 521–526.

43] M.T. Zafarani-Moattar, S. Hamzehzadeh, S. Nasiri, A new aqueous biphasicsystem containing polypropylene glycol and a water-miscible ionic liquid,Biotechnol. Prog. 28 (2012) 146–156.

44] A.M. Azevedo, P.A. Rosa, I.F. Ferreira, M.R. Aires-Barros, Optimisation of aqueoustwo-phase extraction of human antibodies, J. Biotechnol. 132 (2007) 209–217.

45] P.A. Rosa, A.M. Azevedo, S. Sommerfeld, W. Backer, M.R. Aires-Barros, Continu-ous aqueous two-phase extraction of human antibodies using a packed column,J. Chromatogr. B 880 (2012) 148–156.

46] P.A.J. Rosa, A.M. Azevedo, S. Sommerfeld, M. Mutter, M.R. Aires-Barros, W.Bäcker, Application of aqueous two-phase systems to antibody purification:a multi-stage approach, J. Biotechnol. 139 (2009) 306–313.

47] H.-O. Johansson, G. Karlström, F. Tjerneld, C.A. Haynes, Driving forces for phaseseparation and partitioning in aqueous two-phase systems, J. Chromatogr. B711 (1998) 3–17.

48] F. Dismer, S. Alexander Oelmeier, J. Hubbuch, Molecular dynamics simulationsof aqueous two-phase systems: understanding phase formation and proteinpartitioning, Chem. Eng. Sci. 96 (2013) 142–151.

49] P.A. Rosa, A.M. Azevedo, M.R. Aires-Barros, Application of central compositedesign to the optimisation of aqueous two-phase extraction of human anti-bodies, J. Chromatogr. A 1141 (2007) 50–60.

50] Y.X. Guo, J. Han, D.Y. Zhang, L.H. Wang, L.L. Zhou, An ammonium sulfate/ethanolaqueous two-phase system combined with ultrasonication for the separationand purification of lithospermic acid B from Salvia miltiorrhiza Bunge, Ultrason.Sonochem. 19 (2012) 719–724.

51] C. Li, J. Bai, W. Li, Z. Cai, F. Ouyang, Optimization of conditions for bacteriocinextraction in PEG/salt aqueous two-phase systems using statistical experimen-tal designs, Biotechnol. Progr. 17 (2001) 366–368.

52] B. Mokhtarani, R. Karimzadeh, M.H. Amini, S.D. Manesh, Partitioning ofCiprofloxacin in aqueous two-phase system of poly(ethylene glycol) andsodium sulphate, Biochem. Eng. J. 38 (2008) 241–247.

53] R.D. Rogers, H.D. Willauer, S.T. Griffin, J.G. Huddleston, Partitioning of smallorganic molecules in aqueous biphasic systems, J. Chromatogr. B 711 (1998)255–263.

54] H.D. Willauer, J.G. Huddleston, R.D. Rogers, Solute partitioning in aqueousbiphasic systems composed of polyethylene glycol and salt: the partitioningof small neutral organic species, Ind. Eng. Chem. Res. 41 (2002) 1892–1904.

55] H.D. Willauer, J.G. Huddleston, R.D. Rogers, Solvent properties of aqueousbiphasic systems composed of polyethylene glycol and salt characterized bythe free energy of transfer of a methylene group between the phases and by alinear solvation energy relationship, Ind. Eng. Chem. Res. 41 (2002) 2591–2601.

56] A.R. Katritzky, K. Tämm, M. Kuanar, D.C. Fara, A. Oliferenko, P. Oliferenko, J.G.Huddleston, R.D. Rogers, Aqueous biphasic systems. Partitioning of organicmolecules: a QSPR treatment, J. Chem. Inf. Comp. Sci. 44 (2003) 136–142.

57] Chemicalize.org (www.chemicalize.org) by ChemAxon (www.chemaxon.com),Properties Viewer, December, 2013.

58] M.M. Bora, S. Borthakur, P.C. Rao, N.N. Dutta, Aqueous two-phase partitioning ofcephalosporin antibiotics: effect of solute chemical nature, Sep. Purif. Technol.45 (2005) 153–156.

[

gr. A 1361 (2014) 67–76

59] S. Saravanan, J.R. Rao, B.U. Nair, T. Ramasami, Aqueous two-phase poly(ethyleneglycol)–poly(acrylic acid) system for protein partitioning: influence of molec-ular weight, pH and temperature, Process Biochem. 43 (2008) 905–911.

60] F. Mashayekhi, R.Y. Chiu, A.M. Le, F.C. Chao, B.M. Wu, D.T. Kamei, Enhancingthe lateral-flow immunoassay for viral detection using an aqueous two-phasemicellar system, Anal. Bioanal. Chem. 398 (2010) 2955–2961.

61] F. Mashayekhi, A.M. Le, P.M. Nafisi, B.M. Wu, D.T. Kamei, Enhancing the lateral-flow immunoassay for detection of proteins using an aqueous two-phasemicellar system, Anal. Bioanal. Chem. 404 (2012) 2057–2066.

62] P. Schofield, D.M. Mbugua, A.N. Pell, Analysis of condensed tannins: a review,Anim. Feed Sci. Technol. 91 (2001) 21–40.

63] S.A. Oelmeier, C. Ladd Effio, J. Hubbuch, High throughput screening basedselection of phases for aqueous two-phase system-centrifugal partitioningchromatography of monoclonal antibodies, J. Chromatogr. A 1252 (2012)104–114.

64] D.H. Atha, K.C. Ingham, Mechanism of precipitation of proteins by polyethyl-ene glycols. Analysis in terms of excluded volume, J. Biol. Chem. 256 (1981)12108–12117.

65] J. Wu, Z. Wang, W. Lin, S. Chen, Investigation of the interaction betweenpoly(ethylene glycol) and protein molecules using low field nuclear magneticresonance, Acta Biomater. 9 (2013) 6414–6420.

66] H.P.S. Makkar, M. Blümmel, K. Becker, Formation of complexes betweenpolyvinyl pyrrolidones or polyethylene glycols and tannins, and their impli-cation in gas production and true digestibility in in vitro techniques, Brit. J.Nutr. 73 (1995) 897–913.

67] C.R. Mace, O. Akbulut, A.A. Kumar, N.D. Shapiro, R. Derda, M.R. Patton,G.M. Whitesides, Aqueous multiphase systems of polymers and surfactantsprovide self-assembling step-gradients in density, J. Am. Chem. Soc. 134 (2012)9094–9097.

68] B.Y. Zaslavsky, Aqueous Two-phase Partitioning: Physical Chemistry and Bio-analytical Applications, Marcel Dekker, Inc., United States of America, 1994.

69] H.P. van Egmond, Natural toxins: risks, regulations and the analytical situationin Europe, Anal. Bioanal. Chem. 378 (2004) 1152–1160.

70] H. Egmond, R. Schothorst, M. Jonker, Regulations relating to mycotoxins in food,Anal. Bioanal. Chem. 389 (2007) 147–157.

71] M. Bensch, B. Selbach, J. Hubbuch, High throughput screening techniques indownstream processing: preparation, characterization and optimization ofaqueous two-phase systems, Chem. Eng. Sci. 62 (2007) 2011–2021.

72] A.M. Azevedo, P.A. Rosa, I.F. Ferreira, J. de Vries, T.J. Visser, M.R. Aires-Barros,Downstream processing of human antibodies integrating an extraction capturestep and cation exchange chromatography, J. Chromatogr. B 877 (2009) 50–58.

73] G.M. de Medeiros e Silva, D.d.A. Viana Marques, T.S. Porto, J.L.L. Filho, J.A.C.Teixeira, A. Pessoa-Júnior, A.L.F. Porto, Extraction of fibrinolytic proteases fromStreptomyces sp. DPUA1576 using PEG-phosphate aqueous two-phase systems,Fluid Phase Equilib. 339 (2013) 52–57.

74] M. Wiendahl, S.A. Oelmeier, F. Dismer, J. Hubbuch, High-throughput screening-based selection and scale-up of aqueous two-phase systems for pDNApurification, J. Sep. Sci. 35 (2012) 3197–3207.

75] S. Oelmeier, F. Dismer, J. Hubbuch, Molecular dynamics simulations on aqueous

two-phase systems—single PEG-molecules in solution, BMC Biophys. 5 (2012)14.

76] P.P. Madeira, C.A. Reis, A.r.E. Rodrigues, L.M. Mikheeva, B.Y. Zaslavsky, Sol-vent properties governing solute partitioning in polymer/polymer aqueoustwo-phase systems: nonionic compounds, J. Phys. Chem. B 114 (2009) 457–462.